Research Projects

The demand for self-driving features in smart cars has significantly increased in recent years. The first step in decision-making for driverless cars involves scene parsing of road images, which uses semantic segmentation. Deep neural networks have been developed for this purpose, but general-purpose compute processors are slow and energy inefficient. This project aims to design, develop, and implement deep networks and FPGA-ASIC-based hardware architectures and accelerators for semantic segmentation of road images. Hardware accelerators will provide a fast and energy-efficient solution, making this project a promising deployment solution. Additionally, suitable deep neural networks will be developed and trained before hardware-based implementations.